Testing For Cointegration Using Engle-Granger Methodology

Many economic variables having nonstationary character are using for the estimation of a structural equation more than trend stationary ones. The conventional wisdom is to remove a stochastic trend by differencing in all nonstationary variables and then to use them in a regression analysis. In the last years of the 20th century it was recognized that the appropriate way, how to treat nonstationary variables is not so straightforward in a multivariate context. The conception of cointegration was introduced by Engle, R. F. & Granger, C., W., J. (1987) and unroll by others. It was showed that variables integrated of the same order are said to be cointegrated if there exists a linear combination among them that is stationary. This idea is the base of the equilibrium theories involving nonstationary variables. One area of application of equilibrium theories is long-run purchasing power parity which requires that real echange rates r(t) = e(t) + p(t) * -p(t) are stationary. Cointegration offers an alternative method to check this theory, where et denotes log of the nominal price of foreign exchange and p(t) and p(t) *, denote the logs of domestic and foreign price levels, respectively. The aim of the article is to explain the Engle-Granger testing procedure for cointegration of sequences f(t) = e(t) + p(t) * with p(t) or f(t) = log((Skk/Czk)(t) * CPIcz(t)/100)) and p(t) = log(CPIsk/100)(t) during the period 1997:1 to 2002:12.